3 ms·
That's one possibility for the harms it may cause. I think there's a reasonable chance that someone who thinks the idea is laughable will, at some point, ask a
by ben_w 2y ago
That's one possibility for the harms it may cause.
I think there's a reasonable chance that someone who thinks the idea is laughable will, at some point, ask an AI to destroy the world specifically to demonstrate that it can't… actually, that part has already happened*, the questions are: (1) how competent does it have to be at following that instruction to actually do so? (2) What resources would someone have to allow it to access (deliberately or accidentally) to pull that off?
* someone has asked an AI to destroy the world, but I don't know their motivation.
Fortunately for everyone, this one was kinda dumb: https://community.openai.com/t/chaosgpt-an-ai-that-seeks-to-destroy-humanity/160028 https://community.openai.com/t/chaosgpt-an-ai-that-seeks-to-...